How could one recording be at a different speed?

This is a bit of an odd one - I’ve been producing and editing an Actual Play podcast for a little over a year now. We have four players with me as GM, we play the game over Hangouts and everyone records themselves locally using Audacity. After recording everyone exports a .flac file and uploads it so I can collate them and begin to edit.

We sync up at the start using the time-honoured approach of clapping against

In the session I have just been editing, one track kept seeming to slip relative to the others, after a while I realised I wasn’t going crazy and looking at the source material it appears to be very slightly faster than everyone else’s tracks. The effect is slight - trying to correct it with the speed adjustment tool suggests it is probably only about 1.36% faster.

However when I listen to the recording it also seems as though the timing is “slipping” somehow as we go on- this track seems to wander into and out of sync with the others at various points. The pitch seems normal, although these are fairly small fluctuations so it might not show.

I think she is using an older Windows machine ( maybe Win 7 ) but we have not run into this problem before.

Can anyone suggest what might have happened or give any tips on how to keep it from happening again?

It’s always going fast, right? That’s can mean a slow recorder, or a recorder that can’t keep up. I’m going with can’t keep up since the track seems to be wandering by itself.

So good computer hygiene is called for. Make sure it’s not filling up. Make sure only the minimum applications are running, etc.


So this is probably a consequence of the computer not having the capacity to record everything it needs to and essentially lowering the recording bit-rate?

There are (at least) two possibilities:

  1. The clock on the sound card of the one computer is running 1.36% off speed. That is a pretty big error for such things, (more typical would be ~0.01%). So I would not think this likely, but it might be worth testing with a different sound card or USB sound interface if that is possible.

  2. That one computer is indeed dropping samples. Typically that would lead to very noticeable pops at the locations where data is dropped.

One of the complaints is error wandering…


We always go into these things with the idea that there is some operating problem, but it doesn’t have to be. Something could be breaking. Computer systems don’t live forever. But yes. One way to get fast motion is slow down the movie camera. That wandering thing does throw a wrench into the works. That’s quite a bit harder to account for.


Your process is recommended. You can get really lucky with a Skype recording (and it’s certainly fast) but you lose all the production tools. The two downsides are post production time and your specific problem of having one performer crash. You could be stuck troubleshooting across multiple timezones.

First I heard of that time thing for syncing.

I’d be doing something like you clap and everybody holds one earphone up to their microphone.


You’re perfect to confirm a post production rule of thumb, that in general, it takes five times the length of the show to produce it. So if you have an hour show, you’re going to be there for five hours putting it together—again, on average.

For people who poo-poo this idea it is suggested that it takes two times just finding all the errors and then certifying the corrections by playing the final. And we’re not even editing yet.

That’s why one and done productions are so popular. That’s also why I refuse to watch instructional YouTube videos. Most of them should be half their size just by cutting out the mistakes.

That and the hand-held videos shooting a demo that takes two hands…


My experience comes in at a little slower than that in general - on this I’m getting about 10 minutes of output for every hour of editing, but it’s an acceptable price for a show to have production values ( not high production values admittedly :slight_smile: ) and it seems to be gradually building up listening.

I think the most likely cause is definitely the person who has the tracking problem just running into trouble on their PC. I hadn’t realised that could happen, but it makes sense that it might if it is trying to run Audacity, Hangouts and Roll20 while also being a somewhat antique PC.

That’s five times if you’re good at it and the wind is going your way.

Even if nothing is broken, I doubt the super accurate, atomic level time-base generators cost more than four cents, so you can get show drift even if everything is working perfectly.